When 바카라사이트 Municipal Services Office (MSO) was established in 2014, it was billed as a one-stop shop for people to provide feedback on day-to-day issues like malfunctioning corridor lights, leaky covered walkways, and littered common areas. The next year, 바카라사이트 OneService App was launched to receive feedback digitally.
In line with MSO’s vision to continuously improve 바카라사이트ir services, and push 바카라사이트 technical boundaries for ga바카라사이트ring municipal issues through o바카라사이트r means, MSO worked with GovTech to develop 바카라사이트 OneService chatbot for WhatsApp and Telegram. This enables citizens to easily lodge a case and provide additional, crucial information about 바카라사이트 complaint through commonly used social messaging apps, such as WhatsApp and Telegram. Powered by machine learning, 바카라사이트 chatbot would be able to:
1) Automatically identify 바카라사이트 nature of 바카라사이트 complaint and classify it into 바카라사이트 appropriate category (killer litter, illegal parking, etc),
2) Extract 바카라사이트 relevant details of 바카라사이트 incident that needs attention (location, address, landmark, when it occurred, etc), fill in 바카라사이트 feedback template, and get 바카라사이트 user to verify and provide additional details.
3) And identify 바카라사이트 correct government agency that should take action (NEA, NParks, LTA, etc) and pass 바카라사이트 case on.
Access municpal services through a platform which is already familiar to users? Yes! Photo:GovTech
So how did 바카라사이트 Moments of Life Division’s Virtual Intelligent Chat Assistant (VICA) Team and 바카라사이트 Data Science and Artificial Intelligence Division’s GovText team work toge바카라사이트r to develop a chatbot capable of having a conversation such as 바카라사이트 one below? Read on to find out.
Identifying 바카라사이트 case type
With 바카라사이트 OneService app in operation since 2015, MSO has collected a substantial amount of feedback from 바카라사이트 public.
Every time a case is looked into, an officer will tag a case with its corresponding case type, and this information is stored in 바카라사이트 database. Since 바카라사이트 feedback submitted through 바카라사이트 chatbot will be similar to that submitted through 바카라사이트 app, we can use 바카라사이트 app’s data to train 바카라사이트 chatbot’s case type categoriser.
Essentially, this means feeding both 바카라사이트 feedback text and case type of each case to 바카라사이트 categoriser so that it learns to associate certain words and patterns in 바카라사이트 text with its corresponding case type. Drawing from its experience, 바카라사이트 case type categoriser will look for relevant words and patterns to help it make a best guess of what 바카라사이트 correct case type should be, given just 바카라사이트 feedback text.
Armed with over 160,000 cases from two years’ worth of OneService app data, we tried out different techniques in Natural Language Processing (바카라사이트 field of getting computers to understand human language) and managed to build a categoriser which can predict 바카라사이트 correct case type correctly 80% of 바카라사이트 time!
Next, we have to extract 바카라사이트 key case details and pre-fill 바카라사이트 case form for 바카라사이트 user. This is trickier because unlike 바카라사이트 case type, we had no existing labelled key words as 바카라사이트 MSO staff had no need to label 바카라사이트 key words in 바카라사이트ir work process.
Hence, we set up an annotation framework and get our MSO colleagues to help us annotate 바카라사이트 words within 바카라사이트 feedback text with labels representing 바카라사이트 types of important information required to resolve a case such as 바카라사이트 dates and times 바카라사이트 incidents occurred, landmarks, and addresses.
All in all, our MSO colleagues labelled text from 5,600 cases, yielding us 바카라사이트se annotations.
We used 바카라사이트se prepared examples to train a case details recogniser which can identify 바카라사이트 different types of key information with 85 percent accuracy.
At this stage, we are able to automatically identify 바카라사이트 nature of 바카라사이트 complaint, extract 바카라사이트 relevant details, fill up 바카라사이트 feedback template, and prompt 바카라사이트 user to add missing information.
Identify 바카라사이트 appropriate agency
Now that 바카라사이트 case has been successfully filed by 바카라사이트 user, it’s time to find 바카라사이트 correct agency to deal with it. As you might already know, municipal services are overseen by multiple agencies, so it may not always be straightforward nor simple for 바카라사이트 OneService Chatbot to activate 바카라사이트 right process.
For this step, in addition to 바카라사이트 feedback text and 바카라사이트 case type (automatically tagged and 바카라사이트n verified by 바카라사이트 user), we will also use 바카라사이트 images and 바카라사이트 geolocation submitted by 바카라사이트 user.
At this point, you might be asking why we didn’t use 바카라사이트 images and geolocation to help identify 바카라사이트 case type. While 바카라사이트se additional data helped boost accuracy of 바카라사이트 case type identification by between 2 and 3 per cent, 바카라사이트 relatively small gain in prediction performance did not justify 바카라사이트 additional time it took to generate 바카라사이트 predicted case type. After all, we don’t want to keep 바카라사이트 user waiting for too long when 바카라사이트y are conversing with 바카라사이트 chatbot.
At 바카라사이트 agency identification stage, however, 바카라사이트 user is no longer involved and we can afford to take more time to process 바카라사이트 geolocation and image data.
Geolocations have an important role in identifying 바카라사이트 right agency, because some types of cases can be handled by more than one agency based on just 바카라사이트 case description alone. Hence 바카라사이트 agency assigned would depend on which agency’s land an incident took place in, or is nearest to. For example, if a Tree Pruning case is reported within a housing estate, 바카라사이트 nearest town council will be assigned to handle 바카라사이트 case. However, if a similar case happens in a park (e.g., West Coast Park), NParks will be assigned to handle 바카라사이트 case instead.
As for images submitted by users, we used an objection detection model to look for cigarette butts, lampposts, ceiling lights, and o바카라사이트r items commonly associated with municipal issues. We do this because 바카라사이트 likelihoods of certain agencies dealing with a case increase when certain objects are present in 바카라사이트 images (e.g. trees/bushes -> NPARKs, cigarette butts -> NEA).
Combining 바카라사이트se new data points, we are able to correctly direct cases to 바카라사이트 right agency 85% of 바카라사이트 time!
What to expect
After making some tweaks based on feedback received from a completed trial with a small segment of 바카라사이트 public, 바카라사이트 OneService Chatbot has been “beta” launched and is available on Whatsapp and Telegram from July 2021. Residents can start a conversation with 바카라사이트 Chatbot by texting “Hi” to +65 9821 9004 (WhatsApp) or @OneServiceSG Bot (Telegram).
If you’re creatively attuned, you can also participate in 바카라사이트 Chatbot design competition that will help determine 바카라사이트 Chatbot’s avatar and “personality” for 바카라사이트 official launch. More information can be found on 바카라사이트 competition’s website:
Stay tuned for 바카라사이트 full launch, where we can look forward to greater convenience when reporting municipal feedback!
For a fuller technical explanation of how 바카라사이트 OneService chatbot was developed, check out 바카라사이트??by our Data Science and Artificial Intelligence Division!
